As AI tools become more common in people’s daily work, researchers seek to discover its effects on the labor market, especially for workers at the start of the career.
A document from the Stanford Digital Economy Lab, which is part of the Stanford Institute for Man Centered on Man, has now found early evidence that employment has taken up for young workers in the professions that use the generator. Since the widespread adoption of AI tools began at the end of 2022, a split has appeared and software engineers at the start of their career are among the hardest.
Researchers have used data from the largest payroll supplier in the United States, automatic data processing (ADP), to obtain up-to-date data and gain data for millions of workers in industries, locations and age groups. While other data may take months to go out, the researchers published their results at the end of August with data until July.
Although there has been an increase in demand for AI skills on the labor market, the generative AI tools are improving a lot to perform some of the same tasks generally associated with workers at the start of their career. What AI tools do not have is the experiential knowledge acquired over the labor market, which makes posts higher less vulnerable.
These graphs show how employment over time is compared to the start of a career, development and higher workers (all professions). Each age group is divided into five groups, on the basis of AI exposure, and standardized at 1 in October 2022, when popular generative tools of AI have become accessible to the public.
The trend can be a warning sign for more widespread changes, and researchers plan to continue to follow the data. “There may be reversals in these job cuts. Other age groups may become more or less exposed [to generative AI] And have different diagrams in their employment trends. So we will continue to follow this and see what’s going on, ”says Bharat ChandraOne of the authors of the article and a postdoctoral scholarship holder at the Stanford Digital Economy Lab. In the most “exposed” jobs, AI tools can help or do more work that people do daily.
So what does that mean for engineers?
With the rise of AI coding tools, software engineers have been the subject of many discussions, both in the media and research. “There have been contradictory stories about the question of whether this work is affected by AI, especially for entry -level workers,” explains Chandar. He and his colleagues wanted to find data on what is going on now.
Since the end of 2022, software engineers at the start of their career (between 22 and 30 years) have had a drop in employment. At the same time, the use of intermediate and higher level has remained stable or cultivated. This occurs in the work most exposed to AI, and software engineering is an excellent example.
Since the end of 2022, employment for software developers at the start of their career has dropped. However, employment for other age groups has experienced modest growth.
Chandar warns that, for specific professions, the trend may not be motivated by AI alone; Other changes in the technology industry could also cause the decline. However, the fact that he holds in all industries suggests that there is a real effect of AI.
The Stanford team also examined a wider category of “computer professions” based on the American work classifications – which includes material engineers, web developers, etc. – and found similar results.
Employment growth between October 2022 and July 2025 per age and IA exhibition group. The quintiles 1 to 3 represent the lowest AI exposure groups, which have grown from 6 to 13%. Quintiles 4 to 5 are the jobs most exposed to AI; Employment for the youngest workers in these jobs dropped by 6%.
Part of the analysis uses data from the anthropogenic economic index, which provides information on how IA products of anthropic is used, including the estimates of the question of whether the types of requests used for certain professions are more likely to automate work, potentially replace employees or increase the production of an existing worker.
Thanks to this data, the researchers have been able to estimate whether the use by the occupation of the AI is generally complete the work of the employees or replaces them. The jobs in which the tools of AI increased work did not see the same drop in employment, compared to the roles involving tasks that can be automated.
This part of the analysis was based solely on the anthropic index. “Ideally, we would also like to obtain more data on the use of the AI of other AI companies, in particular the Open and Google AI,” explains Chandar. (A recent article by Microsoft researchers noted that the use of co -pilot aligned closely with the EA exposure estimates that the Stanford team used.)
In the future, the team also hopes to extend to employment data outside the United States.
From your site items
Related items on the web
